LINQ
Fast Scalable Feedback
For Quantum Error Correction And Advanced Qubit Operations
LINQ is the deterministic communication network inside the Qblox Cluster. It connects up to 120 Q1 sequence processors per mainframe, allowing measurement data and conditional triggers to be shared between sequencers, within or across modules, with bounded latency and no host computer.
button text
Built for fast, mainframe-wide feedback
< 650 ns FOR ANY CONDITIONAL OPERATION ACROSS THE CLUSTER
SIMULTANEOUS MULTI-QUBIT READOUT
ALL-TO-ALL BRANCHING
REAL-TIME PARAMETER UPDATES ACROSS UP TO 20 MODULES SPACER
Paired with the synchronisation layer SYNQ
LINQ forms the deterministic backbone for feedback-driven quantum operations at scale.
LINQ at a glance:
FEEDBACK LOOP LATENCY
< 650 ns, fully loaded across up to 20 modules with simultaneous readout, all-to-all branching, and real-time parameter updates, entirely in hardware
BACKPLANE THROUGHPUT
2 Gb/s per Cluster, a margin above 25x at code distance 9
DECODER INTERFACE
QECi standard, vendor-agnostic
DEMONSTRATED QEC SYSTEM INTEGRATION
Riverlane Deltaflow 2, code distances 3, 5, 7 to 9, 6.886 µs full-loop latency on Surface Code-17, 11.886 µs on Surface Code-161
HPC/GPU PATH
NVIDIA NVQLink, GPU-accelerated decoding under CUDA-Q
SEQUENCERS PER FEEDBACK NETWORK
Up to 120 Q1 sequence processors per mainframe
MULTI-CLUSTER SCALING
Throughput scaling linearly per added Cluster
Built for real-time decoder integration
QECi
Native support for the QECi standard, the industry interface for connecting control hardware to third-party decoders without custom engineering.
Riverlane Deltaflow 2 QEC system
Demonstrated across code distances 3, 5, 7, and 9, with a measured 6.886 µs on Surface Code-17 to under 12 µs on Surface Code-161
button text
NVIDIA NVQLink
Q1 sequence processors connect to NVQLink, extending LINQ's feedback model to GPU-accelerated decoding under CUDA-Q, opening a path to hybrid quantum-classical co-processiheng.
button text
Open down to the sequencer
LINQ runs at the Q1ASM assembly layer, giving teams direct, low-level access to every sequence processor in the Cluster. Custom feedback logic, conditional branching, and pulse-level tuning can be written and iterated without waiting on a vendor roadmap to expose a new primitive.
Real-time feedback vs. near-time processing
Quantum error correction and other conditional operations must be completed inside the qubit's coherence time. In a control stack, that separates processing into two categories:
Near-time operations
Runs on classical computers between circuits, typically on millisecond timescales.
- Used for tasks such as recalibration and circuit compilation
- Handles much of the control stack’s classical processing
- Suitable for operations that do not need to complete before the qubit decoheres
Near-time operations
Operates within a fixed latency budget determined by the physics of the qubit.
- Runs the complete measure-decode-correct loop in hardware
- No host computer mid-loop
- Avoids the timing variability introduced by a software round trip
Throughput at scale
As quantum systems grow, backplane throughput matters as much as latency. LINQ provides the bandwidth needed to keep feedback scaling with increasing syndrome data volumes.
2 Gb/s
Throughput per Cluster
25x
The requirement for a Surface Code-161
Scale linearly across Clusters
< 650ns
Fast feedback. Scalable throughput.
The same architecture that closes a feedback loop in under 650 ns also scales to support the syndrome data volume of larger quantum error-correction codes.
Where does real-time LINQ make a difference?
A measurement is discriminated on the readout sequencer, sent over LINQ, and used to trigger a conditional action.
Parallel active reset
Five qubits reset in parallel from a Cluster-wide broadcast in under 511 ns end to end, with no isolated-to-parallel latency cliff.
Quantum error correction
Ancilla syndrome data delivered to Riverlane's Deltaflow 2 QEC system via QECi, correction returned to the data qubits, demonstrated at code distances 3, 5, 7, and 9.
Qubit frequency tracking
Closed-loop drift correction via dispersive readout or Ramsey probes with feedback latencies down to 420 ns (configuration dependent).
Calibration via golden-section search
Real-time qubit calibration algorithm that runs on the Q1 sequencer to correct environmental drift and reach higher gate fidelities.
Entanglement heralding
Photon arrivals timestamped at up to 1/128 ns resolution on the QTM, with feedback at 230 ns on the the Low-Latency Path (LLP) or 994 ns on the TDC path.
Conditional branching
Execution path changes dynamically on mid-circuit measurement results, enabling adaptive circuits and protocols like quantum teleportation.
Where does real-time LINQ make a difference?
Qubit Control Module
This is some text inside of a div block.